A Microsoft director of applied science privately called the mass scraping of news for AI training “an astonishing theft of unprecedented proportions” and perhaps “the largest theft of labor in human history,” according to newly unredacted filings in the copyright lawsuit The New York Times and other publishers brought against OpenAI and Microsoft.
The quotes from Brent Hecht’s January 2023 memo appear in a summary judgment motion from the news plaintiffs that was unsealed Thursday, Ars Technica reported. As TechCrunch noted, much of the new material comes from the Times’ own brief rather than the underlying exhibits, which remain sealed, so the quotes appear without their full context.
What the publishers say the documents show
The filing says OpenAI’s mid-training datasets contain more than 91,692 copies of works from the Times, the New York Daily News and the Center for Investigative Reporting, and that a Common Crawl-derived dataset held more than 2 million documents from nytimes.com. It says OpenAI gave Microsoft its entire GPT-3 training dataset to evaluate for Microsoft products, and that Microsoft supplied training data to OpenAI through initiatives called Project Taxi and Project Mango.
The plaintiffs focus on market substitution, a key factor in fair use. Microsoft data showed Copilot cut click-through rates for some news plaintiffs by 83% to 93% compared with traditional Bing search, according to the filing. A January 2024 Microsoft presentation by Hecht described a “doom loop” that would “hurt the performance of our models and the entire web at the same time.” OpenAI’s head of ChatGPT, Nick Turley, wrote that products trained on news posed an “existential threat” to publishers and were “largely substitutive.”
The filing also cites an exchange in which OpenAI researcher Nick Ryder told President Greg Brockman about “a hack” to get around the Times paywall, and Brockman replied, “Ah, nice.” Microsoft CEO Satya Nadella testified in a deposition that “anything that is paywalled should be licensed by anyone who wants to use it,” and said that had he known OpenAI trained on paywalled material, he would have invoked Microsoft’s right to require OpenAI to retrain its models.
Microsoft’s response
A Microsoft spokesperson told Ars that the company’s AI products are a transformative fair use that don’t substitute for news sites, and that Hecht’s documents “reflect one employee’s individual perspective, are not a legal analysis, and do not represent the company’s views.” The spokesperson said Nadella’s testimony addressed “broad principles” and should not be confused with conclusions about the copyright questions before the court. OpenAI did not respond to requests for comment.
Judges have so far been largely receptive to AI companies’ fair use arguments, TechCrunch noted. Steven Lieberman, counsel for the Daily News and seven sister papers, said the documents show “OpenAI and Microsoft knew that what they were doing was wrong.” The publishers say they are ready for trial.
WebProNews is an iEntry Publication